A modelling of the neuro-energetics of Bayesian learning.

确认神经可塑性的能量成本是报酬/成本评估中的一个重要权衡，通过在贝叶斯框架内对经验获得的生物参数进行建模来指导行为和学习。调查贝叶斯学习如何通过最大熵编码等原理与能源效率学习重叠。这将促使人们对代谢高效学习中普遍存在的瓶颈和群体异质性进行调查。这项研究的含义是理解能量成本如何强加信息优先排序的必要条件，从而影响学习和行为倾向。有许多研究概述了阿尔茨海默氏症与氧化应激、代谢障碍和血管交换中的病理变化的关系；因此，了解学习的代谢需求可能会进一步阐明阿尔茨海默氏症等神经退行性疾病的病因。该项目属于EPSRC计算神经科学研究领域。将有效编码假说与贝叶斯大脑假说并列在一起是新颖的。考虑到可塑性的代谢成本作为信息优先排序和决策的一个因素是新的。

Validation of energetic cost of neuroplasticity as an important tradeoff in reward/cost evaluations that instruct behaviour and learning by modelling empirically obtained biological parameters within a Bayesian framework. An investigation into how Bayesian learning overlaps with energy efficient learning through principles such as maximum entropy encoding. This would motivate an inquiry into general bottlenecks and populational heterogeneity in metabolically efficient learning. Implications of this research are an understanding of how energetic cost imposes a requisite for informational prioritisation effecting learning and behavioural tendencies. There is much research outlining how Alzheimers is related to oxidative stress, metabolic dysfunction and pathology in vascular exchange; therefore, understanding the metabolic demands of learning may further elucidate the etiologic of neurodegenerative diseases such as Alzheimers.This project falls within the EPSRC computational neuroscience research area.Juxtaposing the efficient encoding hypothesis with the Bayesian brain hypothesis is novel.Considering the metabolic cost of plasticity as a factor in information prioritisation and decision making is novel.